DE69626349T2 - Frequency estimation using iterative filtering, especially for cellular telephone systems - Google Patents

Description

Field of the invention

The present invention relates to wireless mobile communications, and more particularly relates to carrier frequency determination by mobile receivers using iterative frequency estimation.

Background of the invention

Wireless communications, particularly cellular communications, such as GSM ("the Global System for Mobile communications") or NADC ("North American Digital Cellular Systems"), has become a necessary, if not essential, part of the technology revolution. It has allowed a user to make and receive telephone calls anywhere and at any location, through his mobile cellular telephone or wireless telephone (hereafter referred to as "telephone"), whether he is in a moving automobile slowly making its way through a traffic jam, or in an office building attending a meeting. As long as the user is within a "cell" served by a base station, he is only a telephone call away from his friends, colleagues or business associates.

What should be transparent to the user, for example the seamless connection and accessibility enjoyed by the user, is always very problematic and challenging for the engineers and technologists working behind the scenes. As the user moves through different cell zones, her phone must constantly perform a few tasks. First, the phone should be able to establish a connection with the base station of the serving cell within which the phone is located through a handshake phase. Another task for the phone, after the connection is already established, is to monitor the carrier frequencies of the neighboring cells. Fig. 1 illustrates the relationship between the "serving cell" and the "neighboring cells."

The carrier frequency of the base station, the serving cell, while its frequency is nominally known, cannot be determined exactly when the phone has just moved into a new cell. Therefore, the phone needs to first determine the carrier frequency of the base station of the serving cell, and gradually improve its estimate to within about 100 Hz of the carrier frequency.

In the handshake phase of the phone, such as GSM or NADC, a frequency estimator is generally required to estimate the carrier frequency of a base station for establishing the connection. In a typical TDMA ("Time Division Multiple Access") cellular system, the carrier frequency estimation is usually performed on a frequency correction-burst ("FCB") that repeats every 50 bursts. Due to the existence of strong Gaussian noise, strong co-channel and adjacent-channel interference, and severe attenuation, a conventional frequency estimator, or a tracking unit such as a phase-lock 100p ("PLL"), generally requires 100 or even more FCBs to obtain a reliable frequency estimate. This approach has been very unsatisfactory, as it results in a much longer connection establishment period.

Therefore, it would be desirable to have a more efficient and reliable frequency estimator, as well as a frequency estimation method, for estimating the carrier frequency in the wireless environment.

More in detail, it would be desirable to obtain an accurate frequency estimate in as short a time as possible.

It would also be desirable to obtain a more accurate frequency estimate even in the presence of strong distortions and interference in the wireless environment.

Furthermore, attention is drawn to document WO 92 11706 A which discloses a synchronization process in which a received signal is filtered with an adaptive bandpass filter while the received signal is buffered in memory. The energies of the input signal and the filtered signal are estimated and the gain of the filter is adjusted based on the difference between the energies. The pole of the filter is adjusted so that the frequency of the input signal is centered in the passband of the filter. When a tone is detected, the length of the tone is determined to ensure whether it is a frequency correction burst. If the detected tone is a frequency correction burst, the signal in memory is also the frequency correction burst, which is then filtered in the band-pass filter, and the difference between the frequency of this signal and 67.5 kHz is determined. This difference represents the frequency shift between the carrier frequency of the base station and that of the mobile radio communication device, and can be fed into the local oscillator means to compensate for the frequency shift.

In accordance with the present invention, there is provided a method for estimating a carrier frequency of a base station as set out in claim 1 and an apparatus for estimating a carrier frequency of a base station as set out in claim 8. Preferred embodiments of the invention are disclosed in the dependent claims.

Summary of the present invention

A method and apparatus for iterative carrier frequency estimation of a base station by a mobile station in a wireless communication system is disclosed. Upon detecting a frequency-correction-burst ("FCB") signal from the wireless transmission of the base station, the mobile station first buffers the FCB signal and then filters the buffered signal with a bandpass filter to produce a first set of filtered data. After filtering, the first filtered data is in turn filtered with an auto-regressive filter to produce a corresponding set of second filtered data. With the second filtered data, iterative pole estimation of the AR filter and filtering with the AR filter are performed for a predetermined set of iterations to produce a converged pole estimate, which can then be used to derive the carrier frequency of the base station. The mobile station may also be further adapted to receive and buffer more than one FCB signal to allow averaging of the derived carrier frequencies based on different pole estimates so that a reliable and accurate carrier frequency is obtained.

Short description of the drawing

Additional objects, features and advantages of the present invention will become apparent to those skilled in the art in the following description, in which:

Fig. 1 shows symbolically an area with cellular coverage of different base stations.

Fig. 2 shows a first embodiment of an iterative filtering frequency estimator in accordance with the present invention.

Figure 3 is a flow chart illustrating the method of iterative filtering frequency estimation in accordance with the present invention.

Fig. 4 is a simulation showing the convergence of the estimated frequency a(k) in iterations.

Detailed description of the preferred embodiment

An iterative frequency estimator for estimating the carrier frequency of the base station in a wireless communication environment is disclosed. More in detail, the present invention provides fast and efficient carrier frequency estimation in the wireless environment with strong Gaussian noise, co-channel and adjacent channel interference, and fast channel attenuation. Without using the conventional frequency estimation method, such as the PLL, the present invention uses an iterative filtering method to iteratively process the pre-buffered data and works reliably with one or more FCB bursts.

In the following description, the present invention is described in terms of algorithms and functional block diagrams, which are the usual means for those skilled in the art to communicate with others similarly skilled in the art. It should be recognized by those skilled in the art that the present invention is not strictly limited to its symbolic representation herein, and those skilled in the art can readily modify the illustrated embodiment to implement the present invention for their particular applications.

Referring to Figure 2, an embodiment of the iterative filtering frequency estimator 100 in accordance with the present invention is symbolically shown. From the input end, a band pass filter (BPF) 104 is coupled to an AR ("Auto Regressive") filter 108. The AR filter 108 is then coupled to an AR parameter estimation unit 112 which provides its output to both the AR filter 108 and a frequency calculation unit 116. The output of the frequency calculation unit is provided to a frequency estimation averaging unit 120.

The operation of the frequency estimator 100 will now be described in more detail. As the base station continuously transmits its signals, for example microwave signals, the signals are detected and received by the phones within the cell. The transmitted signals begin with a frequency correction burst (FCB), which is a single tone at a frequency unknown to the phone and which includes 156 samples, "s(n)", at a sampling rate of 270.833 KHz. The FCB samples s (n) are received, detected and buffered by the AFC ("Automatic Frequency Correction") unit of the mobile cellular phone.

The buffered signals s (n) from the FCB burst are first filtered by the BPF 104 to reject out-of-band noise and interference. The BPF 104 therefore produces an output x (n) 106 which is further filtered by a single-pole AR filter 108. Those skilled in the art will recognize that a single-pole AR filter has the following characteristics:

where "a" is an AR parameter or the complex position of the pole and "z" is the variable in the z-transform. The output y(n) 110 of the AR filter 108 can be represented by the following equation:

y(n) = x(n) + a(k) * y(n - 1) Equation 1

where n = 0,1,2, ...155 and a (k) 114 is the AR parameter of the k-th iteration.

After filtering, a (k) is re-estimated by the AR parameter estimation unit 112 according to the following equation:

where k = 1, 2, ...8 and N = total number of samples in each FCB burst. Note that the symbol "*" stands for the conjugate of a variable.

After a(k) has been updated, i.e. the next "k", x(n) 106 is re-filtered by the AR filter 108, and a(k) is again re-estimated using the new y(n) 110. After 8 iterations, for example a(1)...a(8) are obtained as implemented in the current embodiment, a(k) will converge and the frequency estimate f(i) 118 from the current FCB burst can be calculated by the frequency calculation unit 116, according to the following equation:

where F8 represents the sampling frequency and a(k) represents the arctangent of a(k).

In practice, the accuracy of the estimated carrier frequency f(i) is nevertheless very likely to be affected by various factors in the environment, such as adjacent channel and co-channel interference, noise and attenuation. To achieve an even better estimate, the method described above can be applied to one or more than one FCB burst to obtain more than one estimated f(i), as will be appreciated by those skilled in the art.

If more than one f(i) has been calculated, the two estimates with the closest values are averaged 120 to obtain the final estimate "f" 122. If only one FCB burst is used, the frequency estimation process may stop after the frequency calculation unit 116.

Referring now to Fig. 3, a flow chart shows the steps of iterative filtering frequency estimation in accordance with the present invention. First, the data s(n) from the FCB burst is buffered in the phone's memory (201). Then the BPF is applied (preferably an eighth order Butterworth filter) to process the buffered data s(n) or reject out-of-band noise and interference (202).

The single-pole AR filter is used to match the single-tone frequency of the FCB burst (204), by estimating its pole position. Then the pole of the AR filter is re-estimated for the next iteration (206). If the number of iterations is less than 8, the buffered data will be re-filtered with the updated AR filter (208). Otherwise, the carrier frequency f (i) of the base station can be calculated (210).

To obtain optimal estimation when more than one FCB burst is processed, the final estimated frequency will be obtained by averaging the two closest estimates (214). Otherwise, the process can be terminated.

In Fig. 4, the convergence of a(k) after the initial simulation has been demonstrated. Note that a(k) starts at 0 kHz in the first iteration and quickly converges to the target frequency in the eighth iteration.

In an exemplary implementation, the present invention can be used in a GSM handset. The single tone in the FCB burst can be anywhere between 37.7 KHz and 97.7 KHz. The carrier frequency is 900 MHz and the sampling rate is 270 KHz. An eighth order Butterworth filter with the passband from 37.7 KHz to 97.7 KHz is used to process the data (104 and 202). 4 FCB bursts are used to estimate the carrier frequency and take the average (120 and 214). The percentage in which the final result is within 100 Hz of the true frequency was determined to be over 96% in all situations.

Although only a few exemplary embodiments of this invention have been described in detail above, those skilled in the art will readily recognize that many modifications are possible in the exemplary embodiments without materially departing from the novel teachings and advantages of this invention. Accordingly, it is intended that all such modifications be included within the scope of this invention as defined in the following claims.

Claims (10)

1. A method for estimating a carrier frequency of a base station in a wireless communication system using a mobile station, the base station broadcasting wireless signals for the mobile station to detect a time location of a predetermined complex signal within each burst of the wireless signals, the method comprising the following steps: a) buffering (201) the predetermined complex signal from the wireless signals received from the base station to generate buffered data; b) bandpass filtering (202) the buffered data from step (a) with a predetermined bandwidth to generate a predetermined number of first filtered data (106, x(n)). c) filtering (204) the predetermined number of first filtered data (106, x(n)) from step b) with an auto-regressive, hereinafter referred to as AR, filter (108) to generate the predetermined number of second filtered data (110, y(n)) according to the following equation: Y(n) = x(n) + a(k) * Y(n - 1), where x(n) is the first supplied data, y(n) is the second filtered data, and a(k) is an AR parameter of the k-th iteration, where "n" is incremented from 0 to a predetermined number; d) estimating (206) a first pole estimate, a(k + 1) of the AR filter (108) by using the predetermined number of second filtered data y(n) from step c) according to the following equation: where N is the predetermined number of samples within each burst; e) updating (206, 208) the first pole estimate a(k + 1) by repeating "k" iterations from step c) to step d) to obtain a final pole estimate; f) calculating (210) the carrier frequency using the final pole estimate from step e) according to the following equation: where "FS" is a predetermined sampling frequency of the wireless signals, and " a(k)" represents an arctangent of a(k). 2. The method of claim 1, further comprising: g) receiving (212) a plurality of bursts of the predetermined complex signals from the base station to obtain a corresponding plurality of carrier frequencies; h) from the plurality of carrier frequencies, finding a predetermined number of closest carrier frequencies; i) averaging (214) the predetermined number of closest carrier frequencies to estimate a second carrier frequency. 3. A method according to claim 2, wherein four bursts of the predetermined complex signals are received from the base station to obtain four carrier frequencies, and wherein two closest carrier frequencies are averaged. 4. The method according to claim 1, wherein the y(n) in step c) are obtained by means of either a two-pole AR filter (108) or a three-pole AR filter (108). 5. The method of claim 1, wherein the "n" is incremented from 0 to 155 from step c) to step e). 6. The method of claim 1, wherein the predetermined complex signal is a complex tone representative of a frequency correction burst (FCB). 7. The method according to claim 1, wherein step c) to step e) is iterated k = 8 times. 8. An apparatus for estimating a carrier frequency of a base station in a wireless communication system using a mobile station, the base station transmitting wireless signals for the mobile station to detect a time location of a predetermined complex signal (102) within each burst of the wireless signals, the apparatus comprising: Buffer means for buffering the predetermined complex signal (102) from the wireless signals received from the base station to generate buffered data; Bandpass filter means (104) for filtering the buffered data from the buffer means, with a predetermined bandwidth, to generate a predetermined number of first filtered data (x(n), (106)); AR filter means (108) for filtering the predetermined number of the first filtered data (x(n), (106)) from the bandpass means (104) to generate the predetermined number of the second filtered data (y(n), (110)), according to the following equation: y(n) = x (n) + a(k) * y (n - 1), where x(n) is the first filtered data, y(n) is the second filtered data, and a(k) is an AR parameter of the k-th iteration, where "n" is incremented from 0 to the predetermined number; AR parameter estimation means (112) for obtaining, by at least one iteration, a first pole estimate a(k + 1) (114) of the AR filter means (108) by using the predetermined number of second filtered data (y(n), 110)) from the AR filter means (108) according to the following equation: where "N" is the predetermined number of samples within each burst; Frequency calculation means (116) for calculating the carrier frequency using the pole estimate from the AR parameter estimation means (112) after at least one iteration according to the following equation: where "Fs" is a predetermined sampling frequency of the wireless signals, and a(k) represents an arctangent of a(k). 9. Apparatus according to claim 8, wherein the buffering means buffers a plurality of bursts of the predetermined complex signals (102) from the base station and the frequency calculating means (116) calculates a corresponding plurality of carrier frequencies, the apparatus further comprising: Frequency estimation averaging means (120) for averaging a predetermined number of closest carrier frequencies to obtain a final carrier frequency. 10. A device according to claim 8, wherein the y(n) are obtained by either a two-pole AR filter (108) or a three-pole AR filter (108).